
Biased AI in Systems of Education
Designing for Equitable AI in the Classroom
Why This Matters
When an AI is told a student loves classical music versus rap music, it sometimes assigns significantly different scores to the exact same essay. But the bias runs even deeper—AI systems can use more authoritative language with students of color, essentially talking down to them.
This isn’t a glitch—it’s AI amplifying the hidden curriculum of education. Just as schools have long struggled with embedded biases, AI tools are now automating these same inequitable patterns at scale.
Our research-based activities help teachers and students explore these issues through hands-on learning experiences that encourage critical thinking and collaborative discussion.

Beat Bias:
Two identical essays: classical music fan gets 85/100, rap music fan gets 81/100. Same exact words, different scores.
Experience AI Bias in Real-Time
See how AI can show bias when grading student work
AI Bias Explorer
Student A
“in detention yesterday”
Student B
“in honors assembly”
Test Multiple Scenarios
Socioeconomic indicators, language background, and more
Real-Time Results
Watch AI grade differently by context
Explore and Experiment
Create your own bias scenarios
Ready to uncover AI bias?
Try the AI Bias Explorer →Featured Learning Designs
Activities that help students and teachers explore AI bias through collaborative, hands-on experiences.

AI Story Swap: Who Gets to be the Main Character?
Students explore how AI creates stories and who it chooses as the main character. AI often tells stories based on […]

Phone-Free Classroom
Students use ChatGPT and an AI image generator to explore how phone use shapes attention and learning. They compare AI […]

Design Your Own Activity
Want to design an activity that aligns with your current curriculum? We created this Custom GPT to help you think […]

Whose Way Counts? AI and Algebra
Students compare different ways of solving algebra problems—including how AI responds—and reflect on which methods are centered or left out. […]

Next Token, Whose Truth?
This activity helps us see how LLMs work by guessing the next word or token. They use patterns from training […]

Is An AI Picture Worth 1000 Words?
Use this activity to see how AI moves between text and image, including what assumptions it makes.